1. 11 Jul, 2011 1 commit
  2. 05 Jul, 2011 1 commit
  3. 04 Jul, 2011 1 commit
    • Rémi Denis-Courmont's avatar
      Run-time CPU detection for ARM (meaning NEON) · f25efcaf
      Rémi Denis-Courmont authored
      Unfortunately, we cannot emit NEON opcodes when NEON is not explicitly
      enabled (-mfpu=neon), contrary to MMX & SSE on x86. As a consequence,
      this will not work for inline assembler in a non-optimized plugin,
      namely the deinterlacer.
      
      There is also a(n hopefully theoretical) bug whereby the compiler would
      emit NEON instructions in the descriptor or activation callback of a
      NEON plugin. This could then crash if NEON is not supported, even
      before the NEON run-time check is reached.
      f25efcaf
  4. 30 Jun, 2011 1 commit
  5. 25 Jun, 2011 1 commit
  6. 24 Jun, 2011 1 commit
  7. 18 Jun, 2011 1 commit
  8. 14 Jun, 2011 1 commit
  9. 13 Jun, 2011 3 commits
  10. 09 Jun, 2011 1 commit
    • Erwan Tulou's avatar
      skins2: fullscreen support for multiple screens (both Linux and Win32) · d7615a6e
      Erwan Tulou authored
      - by default, fullscreen is performed on the same monitor as the one
        where the video is currently being played back.
      - support for --qt-fullscreen-screennumber if the user wishes to force
        fullscreen on a given monitor.
      
      This fixes trac #4712. Tested on Linux. Tests needed for Windows.
      d7615a6e
  11. 08 Jun, 2011 2 commits
  12. 06 Jun, 2011 1 commit
  13. 04 Jun, 2011 2 commits
  14. 01 Jun, 2011 2 commits
  15. 29 May, 2011 1 commit
  16. 28 May, 2011 1 commit
  17. 26 May, 2011 1 commit
  18. 20 May, 2011 1 commit
  19. 17 May, 2011 1 commit
  20. 15 May, 2011 2 commits
  21. 14 May, 2011 1 commit
  22. 12 May, 2011 1 commit
  23. 11 May, 2011 2 commits
  24. 09 May, 2011 2 commits
    • Rémi Denis-Courmont's avatar
      Do not pass -mmmx or -msse2 to the compiler · 9c61b594
      Rémi Denis-Courmont authored
      x86 GCC does not need those parameters to compile MMX or SSE assembly
      or built-in intrinsics (contrary to ARM GCC w.r.t. NEON). They only
      allow the compiler to issue MMX or SSE instructions for the plain C
      code. We already rely on this tolerant compiler semantic for the CPU
      detection code, and for some optionally accelerated filters (e.g.
      deinterlace).
      
      Disabling MMX and SSE for non-assembly code should have no or
      negligible effects on the affected plugins. On the other hand, it
      ensures that the plugin descriptor can be run by non-MMX/non-SSE CPUs.
      9c61b594
    • Rémi Denis-Courmont's avatar
      Test SSE compiler support on all IA32 architectures, not just 686 · 41346148
      Rémi Denis-Courmont authored
      Same as is done for MMX. Run-time checks determine whether SSE can be
      used in any case.
      41346148
  25. 08 May, 2011 3 commits
  26. 07 May, 2011 2 commits
    • Rémi Denis-Courmont's avatar
      Mark other symbols as hidden · 9887eecc
      Rémi Denis-Courmont authored
      Unfortunately, libtool is yet incapable of doing marking symbols as
      hidden (not exported) at compilation time.  So lets do it manually.
      According to the GCC documentation, this allow some compile-time
      optimizations depending on the platform.
      This was already done by Damien earlier (and removed during the
      plugin headers cleanup).
      
      The check is disabled on Windows, as the compiler accepts the flag,
      but generates lots of warnings during compilation.
      9887eecc
    • Katsushi Kobayashi's avatar
      qtcapture: added support for all QTKit-compatible video input devices · c1cbe147
      Katsushi Kobayashi authored and Felix Paul Kühne's avatar Felix Paul Kühne committed
      includes a couple of tweaks by me and will need further testing and cleaning
      c1cbe147
  27. 03 May, 2011 1 commit
  28. 02 May, 2011 2 commits
    • Rémi Denis-Courmont's avatar
      Move common predefined macros to config.h and common.am · 7af4b7fe
      Rémi Denis-Courmont authored
      Also, __USE_UNIX98 is a libc-internal definition. The build system is
      not supposed to define that directly (we have _GNU_SOURCE).
      7af4b7fe
    • Rémi Denis-Courmont's avatar
      Remove ugly builtin support · 473a276f
      Rémi Denis-Courmont authored
      I don't this abomination to leak through the stable LibVLC interface.
      To link statically, modules should be linked into LibVLC instead.
      
      And then, there is no need to reintroduce the old "builtin" vlc-config
      type if all modules are built the same. In prehistoric times, the
      distinction was needed because some modules were static and some not.
      473a276f